Predictive Controller Design And FPGA Implementation Of Modular Multilevel Converter

The control performance of converter directly affects the driving efficiency of electric vehicle(EV).Modular multilevel converter(MMC)compared with two-level converter has high modularity,low output voltage harmonic,high waveform quality and low switching loss and so on.In particular,topology of MMC with integrated battery keeps the classic MMC’s advantages,besides,motor driving circuit,charging circuit and battery management system(BMS)in EV are integrated by this topology.It greatly improves utilization and reduces weight.Therefore,it is widely concerned and studied.MMC is operated by using switch mode and controlled by on/off mode.It has current tracking,circle current control,sub-module equalization and other control objectives.Metal-Oxide-Semiconductor Field-Effect Transistor(MOSFET)is used as the switch device of MMC,the switching frequency is usually tens to hundreds of kHz,which requires high real-time.Research on the design of finite control set-model predictive control(FCS-MPC)and hardware accelerated implementation in Field-Programmable Gate Array(FPGA).FCS-MPC regards converter as a discontinuous nonlinear actuator.By online select all possible states to find the control variable to minimize the objective function in discrete-time predictive model.Selecting the right objective function allows the control to have more flexibility,and also realizes the purpose of simultaneous optimization of multiple control objectives,and no modulation is required.At the same time,FPGA has programmability and parallel computing structure,which can accelerated to meet the real-time requirements of MMC.The specific work of this paper are:1.Aiming at the control target of three-phase current tracking and circle current suppression of the classic MMC,multi-steps FCS-MPC is studied.First,the discrete prediction model of the system is derived,and then the upper and lower control method are designed.In order to improve the calculation speed and reduce the calculation amount of the upper FCS-MPC algorithm,a simplified optimization method is proposed.At the first step,the optimal and suboptimal control variables are considered,in the next prediction horizon,only the optimal and suboptimal control quantities and the combination of left and right sides are obtained.At the same time,through the lower voltage sort algorithm,the control variable convert from the three-phase switch states to the number of sub-modules in "on" state,reducing the number of control variables and achieving the equalization of the capacitance voltage of the sub-modules.Finally,the effectiveness of the control algorithm is verified by the simulation experiment.2.Aiming at the three-phase current tracking and state of charge(SOC)of battery balance control of the battery integrated MMC,the design of the FCS-MPC is studied.Firstly,the discrete prediction model of the system is derived,and then the upper,middle and lower controllers are designed to achieve the goal of motor drive and battery SOC balance control.The battery SOC balance part is divided into three parts: the bridge arm battery unit SOC balance,the upper and lower bridge arm battery SOC balance and the three-phase battery SOC balance.The upper layer of the control strategy is the circle current controller,which is the active equalization control algorithm of battery SOC.It is composed of several proportional controllers.It is designed by using the relationship between the battery SOC balance of the upper and lower bridge arms of the same phase and the first harmonic component and the DC component of the circle current.The middle layer is FCS-MPC.The motor drive is realized by tracking the phase current.At the same time,the balance control target of circle current controller is realized by tracking the reference voltage provided by the upper controller.The lower layer is sort algorithm,which is a passive balance control algorithm of battery SOC to achieve the SOC balance of bridge arm battery cell.Finally,the simulation experiment of the control system of the battery integrated MMC is completed to verify the effectiveness of the control algorithm.3.In order to improve the real-time performance of FCS-MPC of MMC,the FPGA hardware accelerated implementation is studied.First,compare the two FPGA implementation schemes,then design the software for the two hardware implementation schemes,including the design and verification of floating-point C code,the design and verification of fixed-point C code,and the board level verification of the algorithm.Then compare the resource consumption,operation time and design complexity of the two implementation schemes.The hardware resource consumption of the whole hardware scheme is higher than that of the heterogeneous scheme.The running time of the whole hardware scheme is shorter than that of heterogeneous scheme,but in terms of the complexity of design,the whole hardware scheme is far more complex than that of heterogeneous scheme,so in the end of this paper,heterogeneous scheme is selected for code acceleration,and the operation speed is about 13.4 times faster.Then,the real-time experiment platform of the controller is built,which is composed of MicroAutoBox,ZYNQ and PC.In order to meet the requirement of high-speed data transmission between MicroAutoBox and FPGA in real-time experiment,User Datagram Protocol(UDP)communication is designed,and data transmission accuracy and real-time are verified,through 7 groups of different data receiving and transmitting experiments,verify the accuracy of Ethernet UDP communication,and carry out transmission speed test,the single transmission rate is 12.25 Mbps.Finally,hardware in the loop experiment is carried out,and the experimental results verify the real-time and effectiveness of the algorithm.